1 Maternal LSD1/KDM1A is an essential regulator of and transcription

2 landscapes during zygotic genome activation

3

4

5 Katia Ancelin1,2, Laurène Syx1,3,4, Maud Borensztein1,2, Noémie Ranisavljevic1,2,

6 Ivaylo Vassilev1,3,4 , Luis Briseño-Roa5, Tao Liu6, Eric Metzger7, Nicolas Servant 1,3,4,

7 Emmanuel Barillot 1,3,4, Chong-Jian Chen6 ,Roland Schüle7, and Edith Heard1,2,*.

8

9 1 Institut Curie, 26 rue d’Ulm, Paris 75248, France

10 2 Genetics and Developmental Biology Unit, INSERM U934/CNRS UMR3215, Paris

11 75248, France

12 3 Bioinformatics and Computational Systems Biology of Cancer, INSERM U900, Paris

13 75248, France

14 4 Mines ParisTech, Fontainebleau 77300, France

15 5 HiFiBiO, 10 rue Vauquelin, Paris 75005, France

16 6 Annoroad Technology Co., Ltd, Beijing, China

17 7 Urologische Klinik und Zentrale Klinische Forschung, 79106 Freiburg, Germany

18

19

20 The authors declare that no competing interests exist

21

22

1

23 ABSTRACT

24

25 Upon fertilization, the highly specialised sperm and oocyte genomes are remodelled to

26 confer totipotency. The mechanisms of the dramatic reprogramming events that occur have

27 remained unknown, and presumed roles of modifying enzymes are just starting to be

28 elucidated. Here, we explore the function of the oocyte-inherited pool of a histone H3K4 and

29 K9 , LSD1/KDM1A during early mouse development. KDM1A deficiency results

30 in developmental arrest by the two-cell stage, accompanied by dramatic and stepwise

31 alterations in H3K9 and H3K4 methylation patterns. At the transcriptional level, the switch of

32 the maternal-to-zygotic transition fails to be induced properly and LINE-1 retrotransposons

33 are not properly silenced. We propose that KDM1A plays critical roles in establishing the

34 correct epigenetic landscape of the zygote upon fertilization, in preserving genome integrity

35 and in initiating new patterns of genome expression that drive early mouse development.

2

36 INTRODUCTION

37

38 Gametes are highly differentiated cell types and fertilization of the oocyte by sperm requires

39 major epigenetic remodelling to reconcile the two parental genomes and the formation of a

40 totipotent zygote. In particular, the paternal genome arrives densely packed with protamines

41 rather than , and the maternal epigenome is highly specialised. Maternal factors

42 must unravel these specialised chromatin states to enable zygotic gene activation and

43 development to proceed. Histone tail post-translational modifications (PTMs), and more

44 specifically methylation, appear to be dynamically regulated during this first step of

45 development (Burton and Torres-Padilla, 2010). Histone lysine methylation appears to have

46 different biological read-outs, depending on the modified residue as well as the state of

47 methylation (mono-, di- or tri-). For example, methylation of histone H3 lysine 4 (referred to

48 as H3K4 methylation hereafter) is mainly associated with transcriptionally active chromatin

49 while methylation of histone H3 lysine 9 (referred to as H3K9 methylation hereafter) is

50 usually linked to repressive chromatin. The incorrect setting of some of these histone marks

51 in cloned animals have been correlated with their poor development potential, pointing to

52 their importance during early stages of development (Matoba et al., 2014; Fatima Santos et

53 al., 2003). However, the actors underlying these dynamic changes in histone modifications

54 after fertilization and their impact on the appropriate regulation of zygotic genome function

55 remain open questions.

56 Lysine methylation is tightly regulated by distinct families of conserved enzymes, histone

57 lysine methyltransferases (KMTs), which add methyl groups and histone lysine

58 (KDMs) which remove them (Black et al, 2012). Importantly, KMTs and KDMs show different

59 specificities for their target lysine substrates, as well as for the number of methyl group they

60 can add or remove (from unmethylated -me0-, to dimethylated –me2, and trimethylated -

61 me3; and vice versa).

3

62 Several KMTs and KDMs have been disrupted genetically in model organisms, including

63 mouse, and their loss often leads to lethality or to severe defects in embryogenesis, or else

64 in tissue-specific phenotypes in adults. This has been linked to their important roles in cell

65 fate maintenance and differentiation, as well as in genome stability (Black et al., 2012; Greer

66 and Shi, 2012). However, investigating their potential roles during the first steps of

67 development, after fertilization is frequently hampered by their maternal mRNA and/or

68 pool, which can persist during early embryogenesis and mask the potential impact that the

69 absence of such factors might have (Li et al., 2010). In mice, conditional knock-outs in the

70 female germline that suppress the maternal store of mRNA and protein at the time of

71 fertilization, can be used to examine protein function during the earliest steps of development

72 (de Vries et al., 2000; Lewandoski et al., 1997). In this way, the roles of KMTs and KDMs

73 during early embryogenesis are just starting to be explored. For example, it has been shown

74 that depletion of maternal EZH2 affects the levels of H3K27 methylation in zygotes, although

75 this did not lead to any growth defects during embryonic development (Erhardt et al., 2003;

76 Puschendorf et al., 2008). Another study investigated maternal loss of Mll2 (Mixed lineage

77 leukemia 2), encoding one of the main KMTs targeting H3K4 and revealed its essential role

78 during oocyte maturation and for the embryos to develop beyond the two-cell stage, through

79 gene expression regulation, (Andreu-Vieyra et al., 2010). Importantly, in the presence of

80 maternal EZH2 or MLL2 protein (when wt/- breeders are used), both Ezh2 and Mll2 null

81 embryos die much later in utero (Carroll et al., 2001; Glaser et al., 2006). The roles of these

82 regulators of lysine methylation can thus be highly stage-specific, with very different effects

83 at the zygote, early cleavage or later developmental stages.

84 The LSD1/KDM1A protein (encoded by the gene previously known as Lsd1 but subsequently

85 renamed Kdm1a, which will be the used in this manuscript hereafter) was the first histone

86 KDM to be characterized to catalyse and 2 demethylation and transcriptional

87 repression (Shi et al., 2004). KDM1A was later shown to demethylate H3K9me2 and to

88 activate transcription (Laurent et al., 2015; Metzger et al., 2005). Genetic deletion of murine

4

89 Kdm1a during embryogenesis obtained by mating of heterozygous animals showed early

90 lethality prior to gastrulation (Foster et al., 2010; Macfarlan et al, 2011, Wang et al., 2007;

91 Wang et al., 2009). In light of the above considerations, we set out to study the impact of

92 eliminating or inhibiting the maternal pool of KDM1A during preimplantation development.

93 We report for the first time the crucial role of Kdm1a following fertilization. The absence of

94 KDM1A protein in zygotes derived from Kdm1a null oocytes led to a developmental arrest at

95 the two-cell stage, with a severe and stepwise accumulation of H3K9me3 from the zygote

96 stage, and of H3K4me1/2/3 at the two-cell stage. These chromatin alterations coincide with

97 increased perturbations in the gene expression repertoire, based on single embryo

98 transcriptomes, leading to an incomplete switch from the maternal to zygotic developmental

99 programs. Furthermore, absence of KDM1A resulted in deficient suppression of LINE-1

100 retrotransposon expression, and increased genome damage, possibly as a result of

101 increased LINE-1 activity. Altogether, our results point to an essential role for maternally-

102 inherited KDM1A in maintaining appropriate temporal and spatial patterns of histone

103 methylation while preserving genome expression and integrity to ensure embryonic

104 development beyond the two-cell stage.

5

105 RESULTS

106

107 Depletion of maternal KDM1A protein results in developmental arrest at two-cell stage

108

109 To investigate whether Kdm1a might have a role during early mouse development we first

110 assessed whether the protein was present in pre-implantation embryos using

111 immunofluorescence (IF) and western blotting (Figure 1 A and B). A uniform nuclear

112 localization of KDM1A within both parental pronuclei was observed by IF in the zygote, and

113 at the two-cell stage. The protein was also readily detected by western blot analysis of total

114 extracts of two-cell-stage embryos when compared to nuclear extracts of ESCs. Altogether,

115 these data reveal the presence of a maternal pool of KDM1A.

116 To assess the function of KDM1A in early mouse embryo development, we deleted the

tm1Schüle 117 Kdm1a gene in the female germline during oocyte growth. To this end Kdm1a

cre 118 Zp3 females, carrying a new conditional allele for Kdm1a deletion engineered in the

119 Schüle group (Zhu et al, 2014), and a Zp3 promoter driven cre transgene exclusively

120 expressed in oocytes (Lewandoski et al., 1997) were produced (see also materials and

f/f cre f/f cre 121 methods). These animals are referred as Kdm1a ::Zp3 in this study). Kdm1a ::Zp3

122 females were then mated with wild-type males (Figure 1C). We isolated one- and two-cell

123 stage embryos derived from such crosses to obtain maternally depleted Kdm1a mutant

124 embryos (hereafter named Δm/wt) in parallel to control embryos (hereafter named f/wt) and

125 we confirmed that the KDM1A maternal pool is absent by performing IF (Figure 1A, bottom

126 panels). In parallel, RT-qPCR analysis revealed the absence of Kdm1a mRNA in mutant

127 oocytes (Figure 1-figure supplement 1A).

128

f/f cre 129 Numerous Kdm1a ::Zp3 females were housed with wild-type males for several months,

f/f f/wt 130 however no progeny was ever obtained, in contrast to Kdm1a or females that produced

6

131 the expected range of pup number (4 to 7; data not shown). This indicated that

f/f cre 132 Kdm1a ::Zp3 females are sterile. To determine the possible causes of sterility, control

f/f f/f cre 133 Kdm1a and mutant Kdm1a ::Zp3 females were mated with wild-type males and

134 embryos were recovered on embryonic day 2 (E2) (Figure 1D and E). The total number of

135 oocytes or embryos scored per female was on average 17 for the mutant background (206

136 oocytes or embryos obtained for 12 females studied) and 25 for the control (226 oocytes or

137 embryos obtained for 9 females studied) (see Figure 1D). We found that the proportion of

138 Δm/wt two-cell stage embryos recovered (19%, n=39/206) was far lower than that obtained

f/f cre 139 with f/wt embryos (75% n=170/226) (Figure 1D). Using Kdm1a ::Zp3 females, we also

140 noted a high percentage of fertilized and unfertilized oocytes blocked at meiosis II (MII)

141 (n=95; 46%) compared to those recovered from control females (n=34; 15%). Inspection of

142 control (n=75) and mutant (n=55) MII oocytes revealed a high proportion of misaligned

143 on the metaphase spindle (Figure 1-figure supplement 1B and C) in mutants

144 (41%) compared to controls (17%), suggesting that a lack of maternal KDM1A can lead to

145 segregation defects. Furthermore, upon fertilization, transmission of inherited

146 chromosomal abnormalities was clearly evident, with the frequent presence of micronuclei in

147 KDM1A maternally depleted two-cell embryos (n=40; 63%) (Figure 1-figure supplement 1D).

148 Lastly, 19% (n=39) of mutant embryos were still at the zygote stage compared to 0% in

149 controls (Figure 1D). These results indicate that many MII oocytes lacking germline KDM1A

150 are not competent at ovulation and that when fertilized their first cell cycle is delayed. Similar

151 results were obtained when using females not subjected to superovulation for mating (Fig1-

152 figure supplement 2).

153

154 We next assessed the progress of surviving Δm/wt two-cell embryos by culturing them in

155 vitro. After 24h in culture, 96% of the Δm/wt embryos were found to be arrested at the two-

156 cell stage, unlike f/wt embryos where only 6% showed an arrest (Figure 1D and E). The

157 mutant embryos blocked at the two-cell stage did not progress further in development upon

7

158 prolonged in vitro culture, and eventually fragmented, while the control embryos progressed

159 towards the blastocyst stage (Figure 1E).

160

161 Taken together, these results suggest that the sterility of Kdm1a germline mutant females is

162 in part caused by a severely compromised spindle organization in some oocytes in the

163 second round of meiosis,as well as for the second round of cleavage after fertilization. This

164 immediate loss of viability of the first generation embryos contrasts with the progressive

165 effect seen across generations when spr-5, the Kdm1a homologue in C.elegans is mutated

166 in germline precursors for both gametes (Katz et al, 2009). Also, targeted disruption of

167 Lsd2/Kdm1b, the closest homologue of Kdm1a, in the mouse female germline, was reported

168 to have no effect on oogenesis and early mouse development, but only later at mid-

169 gestation, due to misregulation at some imprinted (Ciccone et al, 2009). Our results

170 show that KDM1B in the female germline is not sufficient to rescue the phenotype of KDM1A

171 maternal depletion after fertilization.

172

173 Inhibition of the enzymatic activity of KDM1A from early zygote stage mimics the

174 maternal deletion phenotype

175

176 The developmental arrest observed at the two-cell stage of Δm/wt embryos could be due to

177 defects carried over by the mutant oocytes, particularly given the chromosome defects

178 observed in a significant proportion of arrested oocytes, and/or to a requirement for KDM1A

179 function after fertilization. To assess a requirement for KDM1A enzymatic activity in early

180 embryos, we tested the impact of KDM1A catalytic inhibition specifically after fertilization. To

181 this end, we treated wild-type zygotes with pargyline, a well-characterized potent chemical

182 inhibitor of KDM1A enzymatic activity (Fierz & Muir, 2012; Metzger et al., 2005) and followed

183 their development in vitro over 48 hours. As shown in Figure 1F, 76% embryos cultured with

184 pargyline were found to be significantly blocked at the two-cell stage and 17% never

185 progressed beyond the 3/4-cell stage. On the contrary, the majority (94%) of mock treated

8

186 embryos developed to the eight-cell stage within 48h, as expected. These data parallel the

187 phenotype of genetic ablation of the KDM1A maternal pool, where 96% of Δm/wt embryos

188 are developmentally arrested at the two-cell stage and 4% at the 3/4-cell stage. Taken

189 together, the genetic depletion and pargyline inhibition data strongly support a requirement

190 for KDM1A enzymatic activity during the zygote and two-cell stage, for embryos to proceed

191 beyond the two-cell stage.

192

193 Abnormal increase of H3K9me3 levels in KDM1A maternally deficient zygotes

194

195 The above observations suggested that the histone demethylase KDM1A plays an important

196 role in early development. At the zygote stage, H3K4 and H3K9 methylation levels appear to

197 be tightly regulated and show highly parental specific patterns (Arney et al., 2002; Lepikhov

198 & Walter, 2004; Santos et al., 2005; Puschendorf et al., 2008; Santenard et al., 2010; Burton

199 & Torres-Padilla, 2010). Given that KDM1A has been implicated in the regulation of H3K4

200 and H3K9 mono and di methylation in previous studies (Shi et al., 2004; Metzger et al, 2005;

201 Stefano et al., 2008; Katz et al., 2009), we investigated whether the methylation levels of

202 these two histone H3 were affected by KDM1A depletion in one-cell stage embryos.

203

204 To this end, we collected f/wt and Δm/wt embryos at embryonic day 1 and analysed them for

205 both H3K4 and H3K9 methylation using specific antibodies against mono (me1), di (me2)

206 and tri (me3) methylation (Figure 2). We used antibodies that show similar patterns in control

207 zygotes to those previously published by others (Arney et al., 2002; Lepikhov & Walter, 2004;

208 Puschendorf et al., 2008; Santenard et al., 2010; Fátima Santos et al., 2005) (see extended

209 experimental procedures). We prioritised single-embryo analysis given the limited amount of

210 material that can be recovered at these early developmental time points, particularly in the

211 context of the depletion of KDM1A (Figure 1D). We first analysed H3K4 methylation patterns

212 (Figure 2A and B). It was previously reported that the paternal pronucleus only gradually

213 shows enrichment in H3K4me2 and me3 during the one-cell stage, while the female

9

214 pronucleus is enriched with these marks from its oocyte origin (Burton & Torres-Padilla,

215 2010; Lepikhov & Walter, 2004). We compared maternal and paternal pronuclear patterns in

216 control and mutant embryos and categorised them according to previously described

217 nomenclature (Adenot et al., 1997). In mid-stage zygotes, the absence of maternal KDM1A

218 does not seem to affect overall H3K4me1, me2 or me3 levels in either the maternal or

219 paternal pronuclei (Figure 2A and B).

220

221 We also assessed whether H3K9 methylation levels were affected in zygotes lacking a

222 maternal pool of KDM1A (Figure 2C and D). H3K9me1 was reported to be equally enriched

223 in both parental pronuclei, while H3K9me2 and me3 are exclusively present in the maternal

224 pronucleus (Arney et al. 2002; Santos et al. 2005; Lepikhov & Walter 2004; Puschendorf et

225 al. 2008; Santenard et al. 2010; Burton & Torres-Padilla 2010). We found that Δm/wt

226 embryos do not seem to differ from f/wt embryos in H3K9me1 levels (Figure 2C and D). In

227 the case of H3K9me2, a complete absence of H3K9me2 staining in the paternal pronucleus

228 was recorded for both control and mutant zygotes. However, we did note a small change in

229 the proportion of embryos displaying H3K9me2 staining in the maternal pronucleus. This

230 suggests that absence of KDM1A may slightly impact on oocyte-inherited H3K9me2 profiles.

231

232 Although KDM1A was shown to specifically induce demethylation of H3K9me1/2 at target

233 genes (Laurent et al., 2015; Metzger et al., 2005), we nevertheless assayed H3K9me3

234 patterns by IF, in case it could also accumulate in absence of KDM1A, due to the presence

235 of specific H3K9 KMT (Cho et al., 2012; Puschendorf et al., 2008). H3K9me3 enrichment is a

236 feature of constitutive heterochromatin, and has been shown to be zygotically enriched at the

237 periphery of nucleolar like bodies (NLBs) within the maternal but not the paternal pronucleus

238 (Burton & Torres-Padilla, 2010; Puschendorf et al., 2008; Santenard et al., 2010) (Figure

239 2C). In Δm/wt zygotes, strikingly elevated levels H3K9me3 were found in the whole maternal

240 pronucleus when compared to controls (grey arrowhead, Figure 2C and D). Even more

241 surprisingly, in Δm/wt zygotes, H3K9me3 could be detected at the periphery of paternal

10

242 NLBs (yellow arrowhead), when compared to controls. Taken together, these observations

243 show that the absence of maternal KDM1A protein results in specifically elevated levels of

244 H3K9me3 in both parental genomes at the zygote stage, and suggest that KDM1A might be

245 engaged with other chromatin modifiers to regulate H3K9me3 immediately after fertilization.

246

247 Abnormal H3K4 and H3K9 methylation patterns after the first cleavage division in

248 KDM1A maternally depleted embryos.

249

250 In order to investigate whether KDM1A activity was important for the regulation of H3K4 and

251 K9 methylation after the first cell cleavage, we examined two-cell stage f/wt and Δm/wt

252 embryos by IF to measure the relative fluorescence intensities (Figure 3). We found that the

253 overall H3K4 methylation levels for mono, di and tri-methylation were significantly elevated in

254 Δm/wt two-cell embryos (Figure 3A), with the most striking effect being seen for H3K4me3

255 where a 6-fold increase was found in mutants compared to controls. Thus, a lack of KDM1A

256 protein has a significant impact on H3K4 methylation levels at the two-cell stage. When

257 H3K9me1, me2 and me3 levels were also examined by IF, we found that all three marks

258 were elevated, with the most significant effect being seen for H3K9me3, which showed a 2.2

259 fold increase in fluorescence intensity particularly at DAPI dense regions of constitutive

260 heterochromatin (Figure 3B).

261

262 To address the specificity of these effects of KDM1A on H3K4 and H3K9 methylation, we

263 tested other histone marks, reported not to be targeted by KDM1A activity. Two such marks,

264 H3K27me3 and H4K20me3, both associated with heterochromatin, were analysed by IF in

265 f/wt and Δm/wt two-cell stage embryos. No significant changes in either of these marks could

266 be detected in mutant compared to control embryos (Figure 3-figure supplement 1 A and B),

267 underlining the specificity of the defects found in KDM1A maternally depleted embryos. As

268 an additional control, we performed IF analysis of two-cell stage embryos generated from

11

269 wild-type zygotes grown for 24hr with pargyline. H3K4me3 and H3K9me3 patterns revealed

270 changes in pargyline-treated when compared to mock-treated embryos (Figure 3-figure

271 supplement 1 C and D). In both, a global increase in staining was detected when compared

272 to controls, although to a slightly lesser extent than in Kdm1a mutant embryos.

273

274 Absence of KDM1A abrogates the normal changes in transcriptome by the two-cell

275 stage

276

277 After fertilization, development initially proceeds by relying on the maternally inherited pool of

278 RNA and protein, followed by massive induction of transcription of the zygotic genome in

279 different waves as shown in Figure 4A. Newly produced transcripts corresponding to zygotic

280 genome activation (ZGA) appear in two phases, first at the zygote stage (corresponding to

281 minor ZGA) and subsequently at the two-cell stage (major ZGA). Transition from the

282 maternal pool to zygotic products is essential for successful developmental progression

283 (Flach et al., 1982). Previous work has shown that KDM1A affects transcription regulation

284 during in vitro embryonic stem cell (ESC) differentiation or during peri- or post-implantation

285 mouse development (Foster et al., 2010; Macfarlan et al., 2001; Wang et al., 2007; Zhu et

286 al., 2014). However, its role has never been evaluated during the very first steps of

287 embryogenesis, when appropriate transcriptional activity is crucial.

288 In the light of our results on chromatin changes described above, and to assess whether

289 transcription might be affected by the lack of the KDM1A maternal pool, we performed IF

290 analysis against PolII and its elongating form (PolIISer2P), which did not reveal any obvious

291 difference between f/wt and Δm/wt two-cell stage embryos (Figure 4-figure supplement 1A).

292 For a direct comprehensive analysis of the transcriptome upon lack of KDM1A, we used RNA

293 sequencing (RNA-seq) for single oocytes and single embryos at the two-cell stage ona

294 cohort of control and mutant samples (Figure 4; Figure 4-figure supplement 2;

295 Supplementary File 1). The method used is based on that of (Tang et al., 2010) which

296 captures poly(A) tail mRNA and allows examination up to 3kb from the 3’ end. The quality of

12

297 our single oocyte or embryo cDNAs was first checked by qPCR for three housekeeping

298 genes (Hprt, Gapdh, Ppia) known to be stably expressed from oocytes to blastocysts (Mamo

299 et al., 2007; Vandesompele et al., 2002). Control and mutant samples displayed similar

300 relative expression for these three genes attesting to the quality of our samples (Figure 4-

301 figure supplement 1C). We next prepared cDNA libraries from single control and mutant

302 oocytes (n=5 each), as well as individual f/wt and Δm/wt embryos (n=8 each) and performed

303 Illumina®-based deep RNA sequencing on these samples (see experimental procedures and

304 analysis for more details; Supplementary File 1). We used DEseq as a normalization method

305 across our samples to assess the relative gene expression between controls and mutants.

306

307 At the two-cell stage, our analysis revealed two sets of genes that become either upregulated

308 (21%; n=2449; FDR=5%) or downregulated (24%; n=2749) in the mutant when compared to

309 control embryos (Figure 4B; Figure 4-figure supplement 2A). Hierarchal clustering based on

310 the transcription profiles showed that all Kdm1a mutant embryos clustered distinctly from the

311 controls (Figure 4C). Furthermore, the analysis of oocyte transcriptomes also revealed that

312 there were fewer genes misregulated in Kdm1a mutant oocytes, than in Kdm1a mutant

313 embryos (Figure 4-figure supplement 2A). Moreover, Principal Component Analysis

314 demonstrated that the gene expression patterns showed greater differences between

315 controls and mutants at the two-cell stage, than in oocytes, and that the two different stages

316 cluster away from each other (Figure 4-figure supplement 2B). The stage comparison also

317 showed that only a subset of genes were misregulated in common, in both oocytes and two-

318 cell stage embryos, upon loss of maternal KDM1A (Figure 4-figure supplement 2C). GO

319 analysis of up or down regulated genes at the two-cell stage (Figure 4F) or oocytes (Figure

320 4-figure supplement 2D) revealed very little overlap in the specific biological functions

321 affected by loss of function of KDM1A before and after fertilization, with the notable exception

322 of cell cycle associated genes. This connects well with the observed phenotype for poor

323 oocyte competence at fertilization and the total developmental arrest at the two-cell stage.

324 These results reveal that absence of maternal KDM1A most likely leads to transcriptome

13

325 changes during oocyte maturation, but to even more serious defects after zygotic gene

326 activation, at the two-cell stage. The latter may be due in part to an aberrant maternal supply

327 of transcripts/, or else to aberrant transcriptional regulation of the zygotic genome in

328 absence of maternal KDM1A.

329 We assessed our two-cell stage RNA-seq data according to the recent Database of

330 Transcriptome in Mouse Early Embryos (DBTMEE) (Park et al., 2013). DBTMEE was built

331 from an ultra-large-scale whole transcriptome profile analysis of preimplantation embryos, in

332 which genes are classified depending on which transcription waves (as in Figure 4A) they

333 are expressed. As shown in Figure 4D (see also Figure 4-figure supplement 3), we assessed

334 the percentage of genes of each of our classes (up; down and not significantly changed) that

335 overlapped with the different DBTMEE categories of transcription switches, from oocyte to

336 two-cell stage. Strikingly, the upregulated genes in Δm/wt embryos fall essentially into the

337 earliest stages and belong to genes annotated as maternal (37% of this category), as minor

338 ZGA genes (39%) and as zygotic-transient (38%). We checked whether the misregulation of

339 these three categories of genes might originate from the oocyte stage changes. We found

340 that only 56 out of 360 of maternal genes, 94 out of 540 of minor ZGA genes and 6 out of 34

341 of 1C transient (Figure4-figure supplement 3 and data not shown) were already upregulated

342 in mutant oocytes. These results reinforce the conclusion that the maternal and zygotic pools

343 of transcripts become more compromised as development proceeds toward the two-cell

344 stage in mutant embryos, rather than being aberrant right from the Kdm1a mutant germline.

345 In clear contrast, the majority of downregulated genes in the Δm/wt were found to belong to

346 the three categories of genes that are normally activated at the two-cell stage, with 50% in

347 the major ZGA class, 37% in the two-cell transient and 50% in the MGA (Mid zygotic gene

348 activation). This suggests that absence of KDM1A compromises the activation of gene

349 expression by the two-cell stage.

350 In order to validate our RNA seq data and the analysis done, we selected four genes with

351 characteristic expression profiles, Atrx (maternal), H2Az (major ZGA), Suv39h1 (2C-

352 transient), Klf4 (MGA), which all encode chromatin associated factors crucial for early mouse

14

353 development. Validation was performed by RT-qPCR in control and mutant oocytes and two-

354 cell embryos. As predicted from our RNA seq results (Supplementary File 2), H2AZ,

355 Suv39h1 and Klf4 failed to be expressed at two-cell stage in Kdm1a mutant embryos (Figure

356 4E). In contrast Atrx which is a known maternal factor, but which is zygotically expressed by

357 the two cell stage, was correctly activated. No difference in expression of Suv39h1, Klf4 and

358 Atrx could be seen between controls and mutants at the oocyte stage, implying that the

359 maternal pool of these mRNAs was not affected by the maternal KDM1A depletion.

360

361 This single embryo transcriptome profiling data reveals an aberrant gene expression profile

362 in Kdm1a mutant embryos, which is likely due to an absence or delay in the transcription

363 switch from maternal-zygote to the two-cell stage pattern for a substantial set of genes (47%;

364 1818 out of 3811 considered; Figure 4-figure supplement 2). Together with the changes in

365 chromatin profiles that we observed at the two-cell stage, we conclude that part of the

366 deficiency in developmental progression could be due to the inappropriate setting of a

367 successful zygotic gene expression program upon KDM1A loss.

368

369 A (GO) analysis of the up-regulated genes classified as maternal to zygote-

370 transient in Figure 4D, revealed a clear over-representation of genes involved in protein

371 transport and localisation as well as contribution to cell cycle (Figure 4F). GO analysis of the

372 downregulated genes from major-to-mid-zygotic activation are implicated in ribosome

373 biogenesis and translation processes (Figure 4F). Collectively, these results suggest that

374 KDM1A is necessary for the transcriptional regulation of specific genetic pathways implicated

375 in fundamental biological functions such as protein production and localisation, and cell cycle

376 regulation. These combined defects could be consistent with the inability of the mutant

377 embryos to develop further than the two-cell stage

378

379 Impact of KDM1A absence on repeat elements, genome integrity and DNA replication

380

15

381 Many transposable elements are known to be expressed in early mouse embryos, as early

382 as zygotic stage, and some of these repeat elements might even be competent for new

383 events of retrotransposition between fertilization and implantation (Fadloun et al., 2013; Kano

384 et al., 2009; Peaston et al., 2004). The repression of some of these transposable elements

385 during preimplantation has been correlated with loss of active chromatin marks such as

386 H3K4me3, rather than acquisition of heterochromatic marks such as H3K9me3 (Fadloun et

387 al., 2013). Interestingly, a previous study using Kdm1a mutant mESCs and late

388 preimplantation embryos found a significant impact on MERVL:LTR repeat expression (for

389 Murine endogenous retrovirus-like LTR), as well as a good correlation for the presence of

390 remnant ERVs within 2kb of the transcription start site of KDM1A-repressed genes

391 (Macfarlan et al., 2011). The increased levels of H3K4me3 and H3K9me3 that we found in

392 Δm/wt two-cell stage embryos and the reported role of KDM1A in late preimplantation

393 embryos prompted us to analyze the effects of maternal KDM1A depletion on repetitive

394 element expression after fertilization. To this end, we investigated our RNA-seq data from

395 control and Kdm1a mutant two-cell embryos for the relative expression of repetitive

396 elements. As our single embryo RNA seq approach was based on polydT priming this

397 restricted our analysis to reads at the 3’ ends of transcripts, which somewhat limited our

398 capacity to detect repeat variation. In particular we could not determine which specific LINE-1

399 families were expressed in the mutants, nor whether the LINE-1 reads we detected

400 corresponded to full-length,and/or intact elements. Nevertheless, our results shows that by

401 far the most abundant categories of expressed repeats at this stage of development were

402 LTRs (long terminal repeat) and non-LTR retrotransposons in f/wt and Δm/wt (95% and 92%,

403 respectively) (Figure 5A). However, no significant impact on expression could be detected in

404 the mutants, with the exception within the non-LTR elements, of quite a significant

405 overrepresentation of LINEs, but not SINEs (for Long/Short Interspersed Nuclear Elements

406 element) (Figure5 A and B). We validated this result by RT-qPCR using individually prepared

407 cDNAs of two-cell stage embryos for three transposable element classes. LINE-1, SINE B1

408 and MuERV-L transcripts are all abundantly expressed in control and mutant embryos, but

16

409 LINE-1 levels show a two-fold increase in the Δm/wt embryos (Figure 5C). No significant up-

410 regulation was seen in ERV-promoter driven genes, that had previously reported to be

411 affected by loss of KDM1A in ESCs (Macfarlan et al., 2011).

412 To further assess the impact that KDM1A depletion has on active LINE-1 transcription, we

413 used a single-cell method, RNA fluorescent in situ hybridization (RNA FISH), which enables

414 the detection of nascent transcripts. We first assessed the quality of our assay by checking

415 the Atrx gene, known to be transcribed at the two-cell stage (Patrat et al., 2009) and

416 expressed at similar levels in mutant and control (Figure 5-figure supplement 1). A

417 comparable proportion of f/wt embryos and Δm/wt two-cell embryos displayed detectable

418 ongoing transcription, as registered by a pinpoint at this locus. Using a probe spanning the

419 full-length LINE-1 element (Chow et al., 2010), we detected LINE-1 RNA in control two-cell

420 stage embryos as displayed by the punctate pattern in nuclei (Fadloun et al, 2013), while

421 RNase-A treated embryos showed no signal (Figure 5D). In the maternally depleted

422 embryos, the arrangement of fluorescent foci appeared extensively modified (Figure 5D).

423 This was confirmed upon analysis of the fluorescence intensity distributions (Figure 5E left)

424 as well as the image composition for the foci (Figure5 E right), which in both cases

425 significantly separated the two types of samples. Our analysis revealed that the active LINE-

426 1 transcription profiles were extensively modified upon the loss of maternal KDM1A.

427 To investigate whether the increase in nascent LINE-1 transcription observed might

428 correspond to full length LINE-1 elements, we assessed by IF for the presence of ORF1, one

429 of the two LINE-1 encoded proteins. At the two-cell stage, we found an approximately 4-fold

430 increase in the proportion of Δm/wt embryos displaying a stronger IF signal, notably in the

431 nucleus (Figure 5F). These results suggest that the LINE-1 deregulation observed at the

432 RNA level might indeed lead to the production and nuclear import of increased levels of

433 LINE-1 ORF1 proteins. We next investigated whether expression of such proteins from

434 transposable elements would have any consequences. We thus performed γH2AX IF

435 staining to assess whether increased DNA damage signalling could be seen in Δm/wt

17

436 compared to f/wt embryos (Figure 5G; Figure 5-figure supplement 2). Half of the mutant

437 embryos displayed a stronger staining for γH2AX, with a significant increase compared to

438 controls (Figure 5G). We also assessed whether this accumulation of γH2AX signals could

439 also be related to replication delays, as reported previously in the case of maternal loss of

440 two components of the polycomb complex PRC1 (Posfai et al, 2012). We performed EdU

441 pulse treatment (a nucleoside analog of thymidine incorporated into DNA) in two-cell

442 embryos, at a stage when they have normally completed S phase (40-41h post hCG

443 injection). This revealed that S phase is delayed in the Δm/wt embryos given the

444 incorporation of EdU in the mutants, while none of the control embryos used in parallel were

445 stained (Figure 5-figure supplement 2). All the mutant embryos delayed in their replication

446 displayed concomitantly intense γH2AX signals. However, 38% of the mutant embryos did

447 not show any EdU incorporation, indicating that they exit S phase, yet, they still show high

448 levels of γH2AX signals. Finally, although, no significant enrichment was directly found for

449 DNA damage pathways when running our GO analysis (Figure4 E), many genes related to

450 DNA damage repair were upregulated (Supplementary File 2). Taken all together, these

451 results suggest that the elevated DNA damage signalling observed could be independent

452 from replication defaults in KDM1A maternally depleted embryos, but might be related either

453 to changes in transcript levels for DNA damage genes or else to the observed increase in

454 LINE-1 activity in Kdm1a mutant embryos at this stage.

18

455 DISCUSSION

456

457 The oocyte stores maternal factors that besides ensuring the first steps of development prior

458 to zygotic genome activation, also enable the epigenetic reprogramming of the parental

459 genomes (Burton and Torres-Padilla, 2010; Li et al., 2010; Messerschmidt et al., 2012;

460 Lorthongpanish et al., 2013; Seisenberger et al., 2013). Although the dynamics of histone

461 modifications have been assessed, the biological relevance of such changes and the

462 identification of the histone modifying enzymes involved in this process are only starting to be

463 identified. In this study, we have focused on the critical function of KDM1A, a histone

464 demethylase for H3K4me1/2 and H3K9me1/me2, that we find acts as a maternal chromatin

465 factor at the time egg fertilization. We show that maternal KO results in abnormal oocytes at

466 the time of ovulation (at meiosis II stage) and prevents development of fertilized eggs beyond

467 the two-cell stage. Similar oocyte defects and developmental block were also observed in the

468 accompanying paper by Wasson et al, where two other independent conditional alleles were

469 used to induce deletion of KDM1A in oocytes, using Zp3-Cre or Gdf9 Cre. In a recent study

470 maternal depletion of KDM1A was found to affect the first division of meiosis and leads to

471 early apoptosis during oocyte growth (Kim et al, 2015). Taken together, these studies show

472 that KDM1A is required during the formation of the female gametes for the two steps of

473 meiosis (Kim et al, 2015; Wasson at al; this study). Our study also reveals that KDM1A is

474 required as a key factor during early post-zygotic embryo development, as the enzymatic

475 inhibition of KDM1A in wild-type embryos resulted in developmental arrest at the two-cell

476 stage, comparable to maternal deletion. We further show that KDM1A is a major regulator of

477 histone H3K4 and H3K9 methylation patterns at the one-two cell stages and that it controls

478 early switches in transcription patterns during development. KDM1A may also have a

479 potential role in the appropriate repression of some LINE-1 retroviral elements.

480

481

482

19

483 KDM1A and the modulation of after fertilization

484

485 H3K4me1/2/3 levels have been shown to increase from the zygote to the two-cell stage,

486 before decreasing again by the four-cell stage (Shao et al., 2014). To date, only one H3K4

487 KMT, MLL2, has been shown to be necessary at the two-cell stage (Andreu-Vieyra et al.,

488 2010). Here, we report that the maternal pool of the KDM, KDM1A, is also necessary at this

489 stage, with its loss leading to global elevation of H3K4me1/2/3. Noticeably, no changes in

490 transcription levels of genes encoding the main H3K4me2/3 KMTs were recorded

491 (Supplementary File 2). This suggests that KDM1A is a key regulator of H3K4 methylation

492 post-fertilization.

493 Moreover, in absence of KDM1A, the transcripts encoding for two main KMTs

494 (SUV39H2/KMT1B and SETDB1/KMT1E) targeting H3K9me1/2 during preimplantation (Cho

495 et al., 2012; Puschendorf et al., 2008) are well detected in two-cell stage mutant embryos

496 (Supplementary File 2). These KMTs could be able to generate H3K9me3 from the excess of

497 H3K9me1/2, produced because of the absence of KDM1A.

498 In conclusion, KDM1A most likely acts in combination with other chromatin regulators in

499 order to keep a tight balance of the global H3K4/K9 methylation levels during early

500 embryonic development.

501

502 KDM1A is involved in the transcriptional switch at the two-cell stage.

503

504 One of the most striking consequences of lack of maternal KDM1A that we observed was the

505 disruption of the wave-like gene expression patterns previously described at the onset of

506 mouse development (Hamatani et al., 2004; Xue et al., 2013). At the two-cell stage, we saw

507 a significant increase in mRNA levels of genes normally expressed maternally or at the

508 zygote stage, and this increase relates more to post-fertilization disruption rather than

20

509 inherited defects from Kdm1a mutant germline. The accompanying manuscript by Wasson et

510 al reports similar findings concerning transcriptional regulation by maternal KDM1A in early

511 stage post fertilization. Maternal and zygotic mRNA excess could reflect a reduced rate of

512 mRNA degradation, maybe related to the developmental arrest, or else the severe

513 impairment of the mutant embryos in the ribosome biogenesis pathways could preclude the

514 translation machinery of their usage and clearance or else a change in the cytoplasmic

515 polyadenylation of the maternal pool of mRNA could also be disturbing their utilization.

516 Lastly, their abundance could also be due, maybe partly, to an increased transcription rate

517 for these genes (and more specifically the one corresponding to the minor ZGA). Given the

518 accumulation of H3K4 methylation that we show in our study for the mutants at this stage

519 and the proven link of this mark with enhanced transcription (Black et al., 2012), we

520 hypothesise that KDM1A might normally be involved in the transcriptional down-regulation of

521 these genes via H3K4 demethylation. Chromatin-based repression is thought to be

522 superimposed on zygotic genome activation and is necessary for the transition from the two-

523 cell to the four-cell stage (Ma et al., 2001; Ma & Schultz, 2008; Nothias et al., 1995;

524 Wiekowski et al., 1997). We propose that KDM1A might be part of such a mechanism, and

525 required for a transition towards two-cell stage specific gene expression patterns (ie in the

526 major ZGA and MGA waves), and therefore for proper development beyond the two-cell

527 stage. The significant absence of the major ZGA and MGA waves in the transcriptome of

528 Kdm1a mutants supports this hypothesis. Whether misplaced or increased H3K9 methylation

529 (Figure 3B) could be involved in failure of transcription activation is not known, but one can

530 speculate that such repressive chromatin and/or absence of KDM1A itself might impair

531 correct recruitment of transcription regulators. So far, a small subset of such factors (TFs and

532 co-regulators) acting at ZGA-gene promoters has recently been suggested to orchestrate the

533 appropriate gene expression patterns following fertilization (Park et al., 2013; Xue et al.,

534 2013). Although, KDM1A was not reported in this study, our results suggest that maternal

535 KDM1A is nonetheless crucial for shaping the transcriptome in early life. Its role in oocyte

536 and embryogenesis may have long lasting effects, as reported in the accompanying paper by

21

537 Wasson et al where a hypomorphic maternal KDM1A, associated with perinatal lethality,

538 showed alterations in imprinted gene expression much later in life. The importance of the

539 maternal pool of KDM1A opens up exciting prospects for the roles of this remarkable histone

540 demethylase in early development.

541

542 KDM1A is instrumental in preserving the genome integrity

543 The control of repeat elements by epigenetic mechanisms, including histone KMTs and

544 KDMs, may be critical in early development. Previous work has suggested that KDM1A may

545 contribute to MERVL element repression in late pre-implantation embryos (Macfarlan et al.,

546 2011). We did not detect any impact on these elements in the Kdm1a mutants immediately

547 post fertilization. However, we did see a small but significant increase in LINE-1 expression

548 and LINE-1 ORF1 protein levels in the Kdm1a mutant embryos. This observation, together

549 with the striking elevation in H3K4me3 levels, is of particular interest in the context of a

550 recent study which proposed that loss of H3K4m3 at LINE-1 elements (rather than a gain in

551 H3K9 methylation) might be critical for their repression during early pre-implantation

552 development (Fadloun et al., 2013). Whether this increase in LINE-1 expression actually

553 leads to an increase in LINE-1 element retrotransposition (ie new insertions) remains to be

554 seen, but the increase in LINE-1 proteins observed in Kdm1a mutant embryos is potentially

555 consistent with such a possibility. In this context, we speculate that misregulation of LINE-1

556 elements in the absence of KDM1A might participate in the early developmental arrest that is

557 observed, via an increased potential of genome instability and activation of some specific

558 DNA damage checkpoints. The increase in γH2AX foci we detected in Kdm1a mutants,

559 independently from replication stalling problems, could also be consistent with this

560 hypothesis. Our results thus support the hypothesis that histone-based defence mechanisms

561 act to safeguard the genome from LINE-1 retrotransposition during preimplantation

562 development, when global DNA hypomethylation might compromises their usual silencing

563 route (Leung & Lorincz, 2012).

22

564 Finally, chromatin status and regulated expression of another family of repeats, located

565 within pericentric heterochromatin, has been proposed to be involved in developmental

566 progression after fertilization, ensuring correct chromosome segregation and

567 heterochromatin propagation (Probst et al., 2010; Santenard et al., 2010). In the absence of

568 KDM1A, we detected aberrant accumulation of H3K9me3 at presumptive pericentric

569 heterochromatin (NLBs) post-fertilization, as well as lagging chromosomes in oocytes, and

570 micronuclei accumulation following fertilisation. Collectively, this data points to maternal

571 KDM1A protein having a potential role at pericentromere/centromere regions that merits

572 future exploration.

573 In conclusion, our findings demonstrate the instrumental role of KDM1A as a maternally

574 provided protein at the beginning of life in shaping the histone methylation landscape and the

575 transcriptional repertoire of the early embryo.

23

576 MATERIALS AND METHODS

577

578 Experimental methods

579 Collection of mouse embryos and in vitro culture. All mice used were handled with care

580 and according to the guidelines from French legislation and institutional policies. Mice

tm1Schüle 581 (Kdm1a ) carrying the targeted mutation allowing the conditional deletion of the first

582 exon of Kdm1a by insertion of two flanking LoxP sites has been engineered and described

583 by R.Schüle group (Zhu et al, 2014). We received mice carrying two copies of this new

tm1Schüle 584 conditional allele Kdm1a , and after transfer in our animal facitilities, they were bred

cre 585 over the well know Zp3 deleter strain which allow CRE mediated recombination

586 specifically in the female germline (Lewandoski et al, 1997). The genetic background of the

f/f cre 587 mice Kdm1a ::Zp3 is a mixture of C57BL/6J and a 129 substrains, and are referred in this

f/f cre 588 manuscript as Kdm1a ::Zp3 mice (as carrying two Kdm1a conditional alleles and a

589 Zp3.cre transgene). To evaluate KDM1A functions during early development, embryos were

f/f cre f/f 590 obtained from superovulated Kdm1a ::Zp3 or Kdm1a females (aged 4–8 weeks) mated

591 with B6D2F1 males (see Figure 1C), and collected in M2 medium (Sigma) at 21-28h (zygote)

592 and 40-42h (two-cell) after hCG (human chorionic gonadotropin) injection. For pargyline

593 treatment (Sigma;1mM final during 24hr) zygotes were in vitro cultured in M16 (Sigma)

594 droplets under mineral oil in a 5% CO2 atmosphere at 37°C. For replication assays, two-cell-

595 stage embryos were collected at 39–40 hphCG, and embryos were cultured in M16 medium

596 1 h, then transferred to M16 containing 50 μM EdU (Click it Life Technologies) for 45 min.

597 Following fixation in 4% PFA for 15 min. permeabilization in PBS 0.5% TritonX-100 for 15

598 min, blocking in PBS 3% BSA., Click-it reaction was performed for 1 h. Washes and new

599 blocking were followed by immunostaining with antibodies against γH2AX (see next section).

600

601 Immunofluorescence staining

24

602 Immunofluorescence was carried out as described previously (Torres-Padilla et al., 2006),

603 with some modifications. After removal of the zona pellucida with acid Tyrode’s solution

604 (Sigma), embryos were fixed in 4% paraformaldehyde, 0,2% sucrose, 0.04% Triton-X100

605 and 0.3% Tween20 in PBS for 15 min at 37°C. After permeabilisation with 0.5% Triton-X100

606 in PBS for 20 minutes at room temperature, embryos were washed in PBStp (0.05% Triton-

607 X100; 1mg/ml polyvinyl pyrrolidone (PVP-Sigma)) then blocked and incubated with the

608 primary antibodies in 1% BSA, 0.05% Triton-X100 for ∼16h at 4°C. Embryos were washed in

609 PBStp twice and blocked 30 minutes in 1% BSA in PBStp and incubated for 2h with the

610 corresponding secondary antibodies at room temperature. After washing, embryos were

611 mounted in Vectashield (Clinisciences) containing DAPI (4',6'-diamidino-2-phénylindole) for

612 visualizing the DNA. Full projections of images taken every 0.5μm along the z axis are shown

613 for all staining, except for the Orf1 for which the middle section is shown only. Antibody

614 staining for H3K4 methylation is in green, while in red for H3K9 methylation and DNA is

615 counterstained with DAPI (blue). For each antibody, embryos were processed identically and

616 analyzed using the same settings for confocal acquisition Staining were repeated

617 independently at least twice. The following antibodies were used (Antibody/Vendor/Catalog

618 #/Concentration): anti-rabbit KDM1A/abcam ab17721/ 1:750, anti mouse

619 H3K4me1/Cosmobio MCA-MBAI0002/ 1:700, anti mouse H3K4me2/Cosmobio MCA-

620 MBAI0003/ 1:700, anti mouse H3K4me3/Cosmobio MCA-MBAI0004/ 1:700, anti-rabbit

621 H3K9me1 kind gift from T.Jenuwein, anti mouse H3K9me2/Cosmobio MCA-MBAI0007/

622 1:500, anti rabbit H3K9me2/ActiveMotif 39239/ 1:800, anti rabbit H3K9me3/ Millipore 07-442/

623 1:200, anti-mouse H3K27me3/ Abcam ab6002/ 1:400, anti-rabbit H4K20me3/ Abcam ab

624 9053/ 1:200, anti-mouse γH2AX/ Millipore 05-623/ 1/200, anti-mouse β-TUBULIN/ Invitrogen

625 32-2600/ 1:1000, anti-mouse POLII CTD4/ Millipore 05-623/1:200, anti-rabbit POLII CTD4

626 S2P/ abcam ab5095/1:200, anti-rabbit ORF1, kind gift from A.Bortvin/ 1:500, Alexa488 goat

627 anti-mouse IgG/ Invitrogen A11029/ 1:500, Alexa568 goat anti-rabbit IgG/ Invitrogen A11036/

628 1:

25

629 Western-blot procedure

630 50 two-cell stage embryos were resuspended in 2-mercaptoethanol containing loading buffer

631 and heated at 85 °C for 15m. SDS-PAGE, Ponceau staining, and immunoblots were

632 performed following standard procedures using a Mini-PROTEAN Tetra Cell System (Bio-

633 Rad). Primary anti-KDM1A (dilution 1:500) and secondary HRP-conjugated goat anti-rabbit

634 (DAKO, Cat.#K4002) were used. 2μg of ESC nuclear extracts were used as control.

635

636 RNA FISH procedure

637 RNA FISH was performed as described (Patrat et al., 2009). Nick translation (Vysis) using

638 Spectrum green or Spectrum red (Vysis) was used to label double stranded probes. The

639 LINE-1 probe used consisted of a full- length Tf element cloned into a Bluescript plasmid as

640 previously described (Chow et al., 2010). The Atrx probe consisted of a BAC (CHORI;

641 reference RP23-260I15). Briefly, embryos were taken at 42h post hCG and the zona

642 pellucida was removed. Embryos were transferred onto coverslips previously coated in

643 Denhardt’s solution, dried down for 30min at room temperature, after all excess liquid was

644 removed. Samples were fixed in 3% paraformaldehyde (pH 7.2) for 10 min at RT and

645 permeabilized in ice-cold PBS 0.5% triton for 1 min on ice and then directly stored in ETOH

646 70°C ethanol at -20°C until processed for RNA FISH. Hybridizations, without Cot1

647 competition, were performed overnight at 37°C in a humid chamber. Excess of probes was

648 eliminated through three washes in 2xSSC at 42°C for 5min each. Slides were mounted in

649 Vectashield containing DAPI.

650

651 Single embryo RNA RT-qPCR and deep sequencing

652 After zona pellucida removal and 3 consecutive washes in PBS-0.1% BSA, single oocytes or

653 whole two-cell stage embryos were transferred into a 0.2ml eppendorf tube (care was taken

654 to add a minimum liquid volume of PBS BSA) and directly frozen on dry ice and stored in -

655 80°C until use. RNA was extracted and amplified as described previously (Tang et al. 2011).

26

656 For quality control and gene expression analysis, quantitative real-time PCR was performed

657 for gene expression on 1/10 dilution of cDNA preparation in 10μl final volume with Power

658 SYBR green PCR master mix (Applied Biosystems) on a ViiA7 apparatus (Life

659 Technologies). The level of gene expression was normalized to the geometric mean of the

660 expression level of Hprt, Gapdh and Ppia housekeeping genes as according to

661 (Vandesompele et al., 2002). For p<0.05 corresponds to * and p<0.001 to ** by t-test. The

662 following primers used in this study are listed as name/ forward primer 5’ to 3’ / reverse

663 primer 5’ to 3’ Hprt/ ctgtggccatctgcctagt / gggacgcagcaactgacatt, Gapdh/

664 ccccaacactgagcatctcc / attatgggggtctgggatgg, Ppia/ ttacccatcaaaccattccttctg /

665 aacccaaagaacttcagtgagagc (as in Duffié et al 2014), Atrx/ tgcctgctaaattctccaca /

666 aggcaagtcttcacagctgt, H2AZ/ acacatccacaaatcgctga / aagcctccaacttgctcaaa, Klf4/

667 agccattattgtgtcggagga/ agtatgcagcagttggagaac, Suv39h1/ ctgggtccacttgtctcagt/

668 ctgggaagtatgggcaggaa, SineB1/ gtggcgcacgcctttaatc / gacagggtttctctgtgtag (Martens et al ,

669 2005), MuERVL/ atctcctggcacctggtatg / agaagaaggcatttgccaga (MacFarlan et al 2011),

670 Kdm1a/ tggagaacacacaatccgga / tgccgttggatctctctgtt, LINE-1 3’UTR/

671 atggaccatgtagagactgcca / caatggtgtcagcgtttgga

672 For RNA deep sequencing, library construction was performed following Illumina

673 manufacturer suggestions. The 26 samples (5 f/f or wt/wt and 5 ∆m /∆m oocytes; 8 f/wt and 8

674 ∆m/wttwo-cell embryos) were sequenced in single-end 49bp reads on an Illumina HiSeq

675 2500 instrument. The depth of sequencing was ranged from 12,500,000 to 35,000,000 with

676 an average around 18,000,000 reads per sample (Supplementary File 1).

677

678 Data procession and analysis

679 Confocal acquisition and image analysis

680 Imaging of embryos following IF and FISH was performed on an inverted confocal

681 microscope Zeiss LSM700 with a Plan apo DICII (numerical aperture 1.4) 63x oil objective. Z

682 sections were taken every 0.4μm (Figure 1 to 3) or 1 μm (Figure 4 and 5). For fluorescence

27

683 intensity measurement on immunofluorescence Z stacks acquisitions, the nuclear area of the

684 stack image was selected, and then the integrated Intensity (intensity divided by the number

685 of voxels represented within the nuclear area) was obtained using the 3D object counter

686 plugin in Image J (Bolte and Cordelière, 2006). For LINE-1 RNA FISH analysis, home-made

687 script for ImageJ were developed that used descriptors defined as (Haralick, R., 1979) to

688 quantitatively study the texture and structure of images (see related manuscript file

689 containing the code in Java text). Distribution of fluorescence intensities or of Haralick

690 parameters (eg entropy) were compared using t-tests, after all data had been tested as

691 belonging to normally distributed populations (Origin8Pro software). For p<0.05 corresponds

692 to * and p<0.001 to **.

693

694 RNA sequencing

695 For the Gene-based differential analysis, quality control was applied on raw data.

696 Sequencing reads characterized by at least one of the following criteria were discarded from

697 the analysis: (more than 50% of low quality bases (Phred score <5); more than 5% of N

698 bases; more than 80% of AT rate At least 30% (15 bases) of continuous A and/or T). Reads

699 passing these filters were then aligned to the mouse mm10 genome using the TopHat

700 software v2.0.6 (Trapnell et al., 2009). Only unique best alignments with less than 2

701 mismatches were reported for downstream analyses. Count tables of gene expression were

702 generated using the RefSeq annotation and the HTSeq v0.6.1 software (Anders et al., 2014).

703 The DESeq R package v1.16.0 (Anders & Huber, 2010) was then used to normalize and

704 identify the differentially expressed genes between control and mutant embryos. Genes with

705 0 counts in all samples were filtered out and only the 60% of the top expressed genes were

706 used for the analysis, as described in the DESeq reference manual. Genes with an adjusted

707 p-value lower than α=0.05 were consider as differentially expressed. Hierarchical clustering

708 analysis for gene expression pattern of 16 libraries was based on Spearman correlation

709 distance and the Ward method, and performed using the hclust function implemented in the

710 gplots v2.16.0 R package.

28

711 In order to study the transposons expression, we performed the mapping of reads passing

712 the quality control using the Bowtie v1.0.0 software (Langmead et al., 2009). This mapping

713 was performed in 2 steps: (i) reads aligned on ribosomal RNA (unique best alignments with

714 less than 3 mismatches in the seed) (GenBank identifiers:18S, NR_003278.3; 28S,

715 NR_003279.1; 5S, D14832.1; and 5.8S, KO1367.1) were discarded (ii) remaining reads

716 were aligned to the mouse mm10 genome, reporting a maximum of 10,000 genomic

717 locations (best alignments without mismatches). Aligned reads were then annotated and

718 intersected with repeats annotation from the repeatMasker database. The transposon counts

719 table was generated using the reads that fully overlap with an annotated repeat and for which

720 all possible alignments are concordant, i.e associated with the same repeat family in more

721 than 95% of cases. The resulting count table was normalized by the total number of reads

722 aligned on repeats. Statistical analysis to identify repeat families with significant changes in

723 expression between control and mutant embryos was performed using the limma R package

724 v3.20.4 (Ritchie et al., 2015). Repeats family with an adjusted p-value lower than α=0.05

725 were consider as significant.

726 The tool AmiGO 2 (Carbon et al., 2009) was used to perform the enrichment Gene Ontology

727 items with the misregulated genes from the Kdm1a mutant two-cell stage embryos.

728

729 Data access

730 The Gene Expression Omnibus (GEO) accession number for the data sets reported in this

731 paper is GSE68139

732

733

734

735 ACKNOWLEDGMENTS

736 We thank A. Bortvin and T.Jenuwein for kind gift of antibodies. We thank Simao Teixeira Da

737 Rocha, Rafael Galupa and Petra Hajkova for critical reading of the manuscript, and members

29

738 of the E.H laboratory for feedback. We acknowledge the pathogen-free barrier animal facility

739 of the Institut Curie, in particular Colin Jouhanneau, and the UMR3215/U934 Imaging

740 Platform (PICT-IBiSA), in particular Olivier Leroy and Nicolas Signolle. MB was supported by

741 the DIM-Stem Pôle, Ile de France, Funding for EH: Equipe labellisée ‘‘La Ligue Contre Le

742 Cancer’’; EU FP7 MODHEP EU grant no. 259743; Labex DEEP (ANR-11-LBX-0044) part of

743 the IDEX PSL (ANR-10-IDEX-0001-02 PSL).

744

745

746 REFERENCES

747 Adenot, P. G., Mercier, Y., Renard, J. P., and Thompson, E. M. (1997). Differential

748 H4acetylation of paternal and maternal chromatin precedes DNA replication and differential

749 transcriptional activity in pronuclei of 1-cell mouse embryos. Development, 124, 4615–25.

750 Anders, S., Pyl T.P., and Huber W. (2015). HTSeq - A Python framework to work with high-

751 throughput sequencing data. Bioinformatics, 5, 166-9. doi: 10.1093/bioinformatics/btu638

752 Anders, S. and Huber, W. (2010). Differential expression analysis for sequence count data,

753 Genome Biology. 11.R106. doi: 10.1186/gb-2010-11-10-r106

754 Andreu-Vieyra, C. V, Chen, R., Agno, J. E., Glaser, S., Anastassiadis, K., Stewart, a F., and

755 Matzuk, M. M. (2010). MLL2 is required in oocytes for bulk histone 3 lysine 4 trimethylation

756 and transcriptional silencing. PLoS Biology, 8. e1000453. doi:10.1371/journal.pbio.1000453

757 Arney, K. L., Bao, S., Bannister, A. J., Kouzarides, T., and Surani, M. A. (2002). Histone

758 methylation defines epigenetic asymmetry in the mouse zygote. Int.J.Dev.Biol, 46, 317–320.

759 Black, J. C., Van Rechem, C., and Whetstine, J. R. (2012). Histone Lysine Methylation

760 Dynamics: Establishment, Regulation, and Biological Impact. Mol. Cell, 48, 491–507.

761 doi:10.1016/j.molcel.2012.11.006

30

762 Bolte, S., and Cordelières, F.P. (2006). A guided tour into subcellular colocalization analysis

763 in light microscopy. J. Microsc. 224, 213–232.

764 Burton, A., and Torres-Padilla, M.-E. (2010). Epigenetic reprogramming and development: a

765 unique heterochromatin organization in the preimplantation mouse embryo. Briefings in

766 Functional Genomics, 9, 444–54. doi:10.1093/bfgp/elq027

767 Carbon S, Ireland A, Mungall CJ, Shu S, Marshall B, Lewis S. (2009). Online access to

768 ontology and annotation data. Bioinformatics. 25, 288-9. doi: 10.1093/bioinformatics/btn615

769 Cho, S., Park, J. S., Kwon, S., and Kang, Y.-K. (2012). Dynamics of Setdb1 expression in

770 early mouse development. Gene Expression Patterns, 12, 213–218.

771 doi:10.1016/j.gep.2012.03.005

772 Chow, J. C., Ciaudo, C., Fazzari, M. J., Mise, N., Servant, N., Glass, J. L., Attreed, M., Avner,

773 P., Wutz, A., Barillot, E., Greally, J.M., Voinnet, O., Heard, E. (2010). LINE-1 activity in

774 facultative heterochromatin formation during X chromosome inactivation. Cell, 141, 956–69.

775 doi:10.1016/j.cell.2010.04.042

776 Ciccone, D.N., Su, H., Hevi, S., Gay, F., Lei, H., Bajko, J., Xu, G., Li, E., and Chen, T.

777 (2009). KDM1B is a histone H3K4 demethylase required to establish maternal genomic

778 imprints. Nature 461, 415–418. doi:10.1038/nature08315

779 De Vries, W. N., Binns, L. T., Fancher, K. S., Dean, J., Moore, R., Kemler, R., and Knowles,

780 B. B. (2000). Expression of Cre recombinase in mouse oocytes: a means to study maternal

781 effect genes. Genesis. 26, 110–2.

782 Duffié R, Ajjan S, Greenberg MV, Zamudio N, Escamilla del Arenal M, Iranzo J, Okamoto I,

783 Barbaux S, Fauque P, Bourc'his D. (2014). The Gpr1/Zdbf2 locus provides new paradigms

784 for transient and dynamic genomic imprinting in mammals. Genes Dev. 28.463-78. doi:

785 10.1101/gad.232058.113

31

786 Erhardt, S., Su, I., Schneider, R., Barton, S., Bannister, A. J., Perez-Burgos, L., Jenuwein, T.,

787 Kouzarides, T., Tarakhovsky, A., and Surani, M. A. (2003). Consequences of the depletion of

788 zygotic and embryonic enhancer of zeste 2 during preimplantation mouse development.

789 Development, 130, 4235–4248. doi:10.1242/dev.00625

790 Fadloun, A., Le Gras, S., Jost, B., Ziegler-Birling, C., Takahashi, H., Gorab, E., Carminci, P.,

791 and Torres-Padilla, M.-E. (2013). Chromatin signatures and retrotransposon profiling in

792 mouse embryos reveal regulation of LINE-1 by RNA. Nat Stuct and Mol Biol, 20. 332-338.

793 doi:10.1038/nsmb.2495

794 Fierz, B., and Muir, T. W. (2012). Chromatin as an expansive canvas for chemical biology.

795 Nat Chem Biol, 8, 417–427. doi: 10.1038/nchembio.938

796 Flach, G., Hjohnson, M., Braude, P. R., Taylor, R. A. S., and Bolton, V. N. (1982). The

797 transition from maternal to embryonic control in the 2-cell mouse embryo. The EMBO J., 1,

798 681–686.

799 Foster, C. T., Dovey, O. M., Lezina, L., Luo, J. L., Gant, T. W., Barlev, N., Bradley, A., and

800 Cowley, S. M. (2010). Lysine-specific demethylase 1 regulates the embryonic transcriptome

801 and CoREST stability. Mol and Cell Biol, 30, 4851–63. doi:10.1128/MCB.00521-10

802 Glaser, S., Schaft, J., Lubitz, S., Vintersten, K., van der Hoeven, F., Tufteland, K. R.,

803 Aasland, R., Anastassiadis, K., Ang, S-A., and Stewart, A. F. (2006). Multiple epigenetic

804 maintenance factors implicated by the loss of Mll2 in mouse development. Development,

805 133, 1423–32. doi:10.1242/dev.02302

806 Greer, E. L., and Shi, Y. (2012). Histone methylation: a dynamic mark in health, disease and

807 inheritance. Nat Rev Genet, 13, 343–357. doi:10.1038/nrg3173

32

808 Hamatani, T., Carter, M. G., Sharov, A. A., and Ko, M. S. H. (2004). Dynamics of Global

809 Gene Expression Changes during Mouse Preimplantation Development. Dev. Cell, 6, 117–

810 131. doi: 10.1016/S1534-5807(03)00373-3

811 Haralick, R.M. (1979). Statistical and structural approaches to texture. Proc. of IEEE. 67786–

812 804.

813 Kano, H., Godoy, I., Courtney, C., Vetter, M. R., Gerton, G. L., Ostertag, E. M., and

814 Kazazian, H. H. (2009). L1 retrotransposition occurs mainly in embryogenesis and creates

815 somatic mosaicism. Genes and Dev, 23, 1303–1312. doi:10.1101/gad.1803909

816 Katz, D. J., Edwards, T. M., Reinke, V., and Kelly, W. G. (2009). A C. elegans LSD1

817 demethylase contributes to germline immortality by reprogramming epigenetic memory. Cell,

818 137, 308–20. doi:10.1016/j.cell.2009.02.015

819 Kim J, Singh AK, Takata Y, Lin K, Shen J, Lu Y, Kerenyi MA, Orkin SH, Chen T.(2015).

820 LSD1 is essential for oocyte meiotic progression by regulating CDC25B expression in mice.

821 Nat Commun, 6:10116. doi: 10.1038/ncomms10116.

822 Langmead, B., Trapnell, C., Pop, M., Salzberg, S.L. (2009) Ultrafast and memory-efficient

823 alignment of short DNA sequences to the . Genome Biol, 10:R25. doi:

824 10.1186/gb-2009-10-3-r25

825 Laurent, B., Ruitu, L., Murn, J., Hempel, K., Ferrao, R., Xiang, Y.,Liu, S., Garcia, B.A.,Wu, H.,

826 Wu, F., Steen, H., and Shi, Y. (2015). A Specific LSD1/KDM1A Isoform Regulates Neuronal

827 Differentiation through H3K9 Demethylation. Mol Cell, 57, 957–970.

828 doi:10.1016/j.molcel.2015.01.010

829 Lepikhov, K., and Walter, J. (2004). Differential dynamics of histone H3 methylation at

830 positions K4 and K9 in the mouse zygote. BMC Dev Biol, 4, 12. doi:10.1186/1471-213X-4-12

33

831 Lewandoski, M., Wassarman, K. M., and Martin, G. R. (1997). Zp3-cre, a transgenic mouse

832 line for the activation or inactivation of loxP-flanked target genes specifically in the female

833 germ line. Curr Biol, 7, 148–51. doi:10.1016/S0960-9822(06)00059-5

834 Li, L., Zheng, P., and Dean, J. (2010). Maternal control of early mouse development.

835 Development, 137, 859–70. doi:10.1242/dev.039487

836 Lorthongpanish, C., Cheow, L.F., Balu, S., Quake, S.R.,Knowles, B.B., Burkholder,

837 W.F.,Solter, D., Messerschmidt, D.M. (2013) Single-cell DNA methylation analysis revealds

838 epigenetic chimerism in preimplantation embryos. Science, 341, 1110-2. doi:

839 10.1126/science.1240617

840 Ma, J., Svoboda, P., Schultz, R. M., and Stein, P. (2001). Regulation of zygotic gene

841 activation in the preimplantation mouse embryo: global activation and repression of gene

842 expression. Biol of Reprod, 64, 1713–21. doi:10.1095/biolreprod64.6.1713

843 Ma, P., and Schultz, R. M. (2008). 1 (HDAC1) regulates histone

844 acetylation, development, and gene expression in preimplantation mouse embryos. Dev Biol,

845 319, 110–120. doi:10.1016/j.ydbio.2008.04.011

846 Macfarlan, T. S., Gifford, W. D., Agarwal, S., Driscoll, S., Lettieri, K., Wang, J., Andrews,

847 S.E., Franco, L., Rosenfeld, M.G., Ren, B., and Pfaff, S. L. (2011). Endogenous retroviruses

848 and neighboring genes are coordinately repressed by LSD1/KDM1A. Genes and Dev, 25,

849 594–607. doi:10.1101/gad.2008511

850 Mamo, S., Gal, A. B., Bodo, S., and Dinnyes, A. (2007). Quantitative evaluation and selection

851 of reference genes in mouse oocytes and embryos cultured in vivo and in vitro. BMC Dev

852 Biol, 7, 14. doi:10.1186/1471-213X-7-14

34

853 Martens JH, O'Sullivan RJ, Braunschweig U, Opravil S, Radolf M, Steinlein P, Jenuwein T.

854 (2005). The profile of repeat-associated histone lysine methylation states in the mouse

855 epigenome. EMBO J. 24.800-12.

856 Matoba, S., Liu, Y., Lu, F., Iwabuchi, K. A., Shen, L., Inoue, A., and Zhang, Y. (2014).

857 Embryonic Development following Somatic Cell Nuclear Transfer Impeded by Persisting

858 Histone Methylation. Cell, 159, 1–12. doi:10.1016/j.cell.2014.09.055

859 Messerschmidt, D.M.,.de Vries, W., Ito, M., Solter, D., Ferguson-Smith, A., Knowles, B.B.

860 (2012). Trim28 is required for epigenetic stability during mouse oocyte to embryo transition.

861 Science, 335; 1499-502. doi: 10.1126/science.1216154.

862 Metzger, E., Wissmann, M., Yin, N., Müller, J. M., Schneider, R., Peters, A. H. F. M.,

863 Gunther, T., Buettner, R., and Schüle, R. (2005). LSD1 demethylates repressive histone

864 marks to promote androgen-receptor-dependent transcription. Nature, 437, 436–9.

865 doi:10.1038/nature04020

866 Nothias, J. Y., Majumder, S., Kaneko, K. J., and DePamphilis, M. L. (1995). Regulation of

867 gene expression at the beginning of mammalian development. The Journal of Biological

868 Chemistry, 270, 22077–80.

869 O’Carroll, D., Erhardt, S., Pagani, M., Barton, Sheila, C., Surani, M. A., and Jenuwein, T.

870 (2001). The Polycomb -Group Gene Ezh2 Is Required for Early Mouse Development. Molr

871 and Cell Biol, 21, 4330–4336. doi:10.1128/MCB.21.13.4330

872 Park, S.-J., Komata, M., Inoue, F., Yamada, K., Nakai, K., Ohsugi, M., and Shirahige, K.

873 (2013). Inferring the choreography of parental genomes during fertilization from ultralarge-

874 scale whole-transcriptome analysis. Genes and Development, 27, 2736–48.

875 doi:10.1101/gad.227926.113

35

876 Patrat, C., Okamoto, I., Diabangouaya, P., Vialon, V., Le, P., Chow, J., and Heard, E. (2009).

877 Dynamic changes in paternal X-chromosome activity during imprinted X-chromosome

878 inactivation in mice. Proc.Natl.Acad.Sci.USA, 97, 1212-1217. doi:10.1073/pnas.0810683106

879 Peaston , A., Evsikov, A., Graber, J., de Vries, W., Holbrook, A., Solter, D., Knowles, B.

880 (2004). Retrotransposons regulate host genes in mouse oocytes and preimplantation

881 embryos. Dev Cell, 4, 597-606. 10.1016/j.devcel.2004.09.004.

882 Posfai, E.,Kunzmann, R.,Brochard, V,Salvaing, J.,Cabuy, E.,Roloff, T.C.,Liu, Z.,Tardat,

883 M.,van Lohuizen, M.,Vidal, M.,Beaujean, N.,Peters, A.H. (2012) Polycomb function during

884 oogenesis is required for mouse embryonic development. Genes Dev. 26:920-32. doi:

885 10.1101/gad.188094.112.

886 Probst, A. V, Okamoto, I., Casanova, M., El Marjou, F., Le Baccon, P., and Almouzni, G.

887 (2010). A strand-specific burst in transcription of pericentric satellites is required for

888 chromocenter formation and early mouse development. Dev. Cell 19, 625–638.

889 doi:10.1007/s00412-007-0106-8

890 Puschendorf, M., Terranova, R., Boutsma, E., Mao, X., Isono, K., Brykczynska, U., Kolb, C.,

891 Otte, A.P., Koseki, H., Orkin, S., van Lohuizen, M., and Peters, A. H. F. M. (2008). PRC1 and

892 Suv39h specify parental asymmetry at constitutive heterochromatin in early mouse embryos.

893 Nat Genet, 40, 411–20. doi:10.1038/ng.99

894 Ritchie ME., Phipson B., Wu D., Hu Y., Law CW., Shi W. and Smyth, G.K. (2015). Limma

895 powers differential expression analyses for RNA-sequencing and microarray studies.

896 NucAcids Res. 43:e47. doi: 10.1093/nar/gkv007

897 Santenard, A., Ziegler-Birling, C., Koch, M., Tora, L., Bannister, A. J., and Torres-Padilla, M.-

898 E. (2010). Heterochromatin formation in the mouse embryo requires critical residues of the

899 histone variant H3.3. Nat Cell Biol, 12, 853–62. doi:10.1038/ncb2089

36

900 Santos, F., Peters, A. H., Otte, A. P., Reik, W., and Dean, W. (2005). Dynamic chromatin

901 modifications characterise the first cell cycle in mouse embryos. Dev. Biology, 280, 225–36.

902 doi:10.1016/j.ydbio.2005.01.025

903 Santos, F., Zakhartchenko, V., Stojkovic, M., Peters, A. H. F. M., Jenuwein, T., Wolf, E.,

904 Reik, W., and Dean, W. (2003). Epigenetic marking correlates with developmental potential

905 in cloned bovine preimplantation embryos. Curr Biol, 13, 1116-1121. doi:10.1016/S

906 Seisenberger, S., Peat, J. R., Hore, T. a, Santos, F., Dean, W., and Reik, W. (2013).

907 Reprogramming DNA methylation in the mammalian life cycle: building and breaking

908 epigenetic barriers. Philosophical Transactions of the Royal Society of London., 368,

909 20110330. doi:10.1016/j.ceb.2013.02.013

910 Shao, G.-B., Chen, J.-C., Zhang, L.-P., Huang, P., Lu, H.-Y., Jin, J., Gong, A-H., and Sang,

911 J.-R. (2014). Dynamic patterns of histone H3 lysine 4 methyltransferases and demethylases

912 during mouse preimplantation development. In Vitro Cell and Dev Biol. Animal. 50, 603-613.

913 doi:10.1007/s11626-014-9741-6

914 Shi, Y., Lan, F., Matson, C., Mulligan, P., Whetstine, J. R., Cole, P. A., Casero, R.A., and

915 Shi, Y. (2004). Histone Demethylation Mediated by the Nuclear Amine Oxidase Homolog

916 LSD1 , Nature. 119, 941–953. doi:10.1016/j.cell.2004.12.012

917 Tang, F., Barbacioru, C., Nordman, E., Li, B., Xu, N., Bashkirov, V. I., Lao, K., and Surani, M.

918 A. (2010). RNA-Seq analysis to capture the transcriptome landscape of a single cell. Nature

919 Protocols, 5, 516–35. doi:10.1371/journal.pone.0021208

920 Torres-Padilla, M.-E., Bannister, A.J., Hurd, P.J., Kouzarides, T., and Zernicka-Goetz, M.

921 (2006). Dynamic distribution of the replacement histone variant H3.3 in the mouse oocyte

922 and preimplantation embryos. Int. J. Dev. Biol. 50, 455–461.

37

923 Trapnell, C., Pachter, L., and Salzberg, S.L.. (2009). TopHat: discovering splice junctions

924 with RNA-Seq., Bioinformatics. 25, 1105-11. doi: 10.1093/bioinformatics/btp120

925 Wang, J., Hevi, S., Kurash, J. K., Lei, H., Gay, F., Bajko, J., Su , H., Sun, W., Chang, H., Xu,

926 G., Gaudet, F., Li, E., and Chen, T. (2009). The lysine demethylase LSD1 (KDM1) is required

927 for maintenance of global DNA methylation. Nat Genet, 41, 125–9. doi:10.1038/ng.268

928 Wang, J., Scully, K., Zhu, X., Cai, L., Zhang, J., Prefontaine, G. G., Krones, A., Ohgi, K.A.,

929 Zhu, P., Garcia-Bassets, I., Liu, F., Taylor, H., Lozach, J., Jayes, F.L., Korach, K.S., Glass,

930 C.K., Fu, X.D.,and Rosenfeld, M. G. (2007). Opposing LSD1 complexes function in

931 developmental gene activation and repression programmes. Nature, 446, 882–7.

932 doi:10.1038/nature05671

933 Wiekowski, M., Miranda, M., Nothias, J. Y., and DePamphilis, M. L. (1997). Changes in

934 histone synthesis and modification at the beginning of mouse development correlate with the

935 establishment of chromatin mediated repression of transcription. J of Cell Science, 110,

936 1147–58.

937 Xue, Z., Huang, K., Cai, C., Cai, L., Jiang, C., Feng, Y., Liu, Z., Zng, Q., Cheng, L., Liu, J-Y.,

938 Horvath, S., and Fan, G. (2013). Genetic programs in human and mouse early embryos

939 revealed by single-cell RNA sequencing. Nature, 500, 593–7. doi: 10.1038/nature12364

940 Zhu, D., Hölz, S., Metzger, E., Pavlovic, M., Jandausch, A., Jilg, C., Galgoczy, P., Herz, C.,

941 Moser, M., Metzger, D., Gunther, T., Arnold, S.J., and Schüle, R. (2014). Lysine-specific

942 demethylase 1 regulates differentiation onset and migration of trophoblast stem cells. Nat

943 Comm, 5, 3174. doi: 10.1038/ncomms4174

944

38

945 Figure titles and legends

946

947 FIGURE 1. Kdm1a maternally deleted embryos arrest at two-cell stage.

948 (A) Immunofluorescence using anti-KDM1A antibody (red) at the zygote and two-cell stage

949 shows nuclear accumulation of KDM1A in control embryos (top). Cre-mediated deletion of

950 Kdm1a in maternal germline (bottom) leads to depletion of the protein after fertilization.

951 Paternal pronucleus (p), maternal pronucleus (m) and polar body (pb) are indicated. DNA is

952 counterstained by DAPI (blue). (B) western blot analysis (left panel) for ESC (lane1) and two-

953 cell stage embryo extracts (lane 2) using anti-KDM1A antibody. Ponceau staining (right

954 panel) is shown as loading control. Molecular weights (kDa) are indicated on the left. (C)

955 Mating scheme and experimental outcomes for the different developmental stages used in

956 this study: f/wt control embryos are obtained from superovulated Kdm1af/f females mated

957 with wild-type males, while Δm/wt mutant embryos are obtained from superovulated

f/f cre 958 Kdm1a ::Zp3 females crossed with wild-type males. (D) Distribution of developmental

959 stages found in f/wt and Δm/wt embryos collected at embryonic day 2 (E2) (expected two-cell

960 stage) and after 24 h of in vitro culture. Females and oocytes/embryos numbers, and the

961 ratio of oocytes/embryos per female, are shown under the graph. See also Figure 1-figure

962 supplement 1 for developmental stage distribution using natural matings without

963 superovulation for females and Figure 1-figure supplement 2 for oocyte analysis. (E) Bright

964 field images representative for two consecutive days of in vitro culture for f/wt and Δm/wt

965 embryos collected at E2. (F) Phenotypes and distribution of developmental stages obtained

966 after 48 h treatment in vitro culture with a catalytic inhibitor (pargyline) of KDM1A in wild-type

967 zygotes recovered at 17 h post hCG injection. Scale bars represent, 10μm and 50μm, in A

968 and D, respectively.

969

970

39

971 FIGURE 2. H3K9me3 heterochromatin levels are defined by maternally inherited

972 KDM1A at the zygote stage.

973 (A and C) IF using antibodies against me1, me2 and me3 of (A) H3K4 (in green) and (C)

974 H3K9 (in red) during zygotic development. Mid to late f/wt and Δm/wt zygote are shown.

975 Paternal pronucleus (p), maternal pronucleus (m) and the polar body (pb) are indicated when

976 present. DNA is counterstained with DAPI (blue). In C, note that in Δm/wt zygotes, H3K9me3

977 is increased in the maternal pronucleus (grey arrowhead) and is localized de novo in the

978 paternal pronucleus (yellow arrowhead). (B and D) Classification of embryos based on

979 staining intensity scores for H3K4/K9me1/2/3) in the paternal versus maternal pronuclei in

980 zygotes. Note that concerning H3K9me2, 50% of Δm/wt embryos have a strong staining

981 versus 35% in controls (which are also up to 20% with no IF signal). The most striking and

982 only significant differences in proportions are seen for H3K9me3 both in maternal (grey

983 arrowheads) and paternal (yellow arrowheads) pronuclei, with p<0.05 using a Chi square

984 test. The scoring is as follows: light grey for no signal; medium green/red for moderate signal

985 and dark green/red for strong signal. Number of embryos and their genotypes are indicated

986 at the bottom of the graph. Scale bar in A and C represent 10μm.

987

988 FIGURE 3. Two-cell stage H3K4 and H3K9 methylation levels are altered upon absence

989 of maternal KDM1A.

990 Immunofluorescence stainings of two-cell stage embryos using antibodies against me1, me2

991 and me3 of H3K4 (A; in green) and H3K9 (B; in red) were performed on f/wt (left panels) and

992 Δm/wt embryos (right panels). Control and mutant samples were processed in parallel and

993 acquired using similar settings at the confocal microscope. DNA is counterstained with DAPI

994 (blue). Projections of z-stacks are shown of representative embryos for each staining. Scale

995 bars, 10μm. Error bars represent S.E.M. By t-test; p<0.05 corresponds to * and p<0.001 to **

996 as performed on the number of embryos indicated below each picture.

40

997 Below each image are shown the relative quantifications for IF intensity levels of me1, me2

998 and me3 of Δm/wt (in black) relative to f/wt (in white) in two-cell stage embryos. Note that no

999 alteration for H3K27me3 or H4K20me3 could be detected for mutant two-cell stage embryos

1000 (Figure 2-figure supplement 1A and B). Also, IF for pargyline-treated two-cell stage embryos

1001 revealed changes in both H3K4me3 and H3K9me3 patterns (Figure 2-figure supplement 1C

1002 and D).

1003

1004 FIGURE 4. Abnormal ZGA upon absence of KDM1A revealed by transcriptome analysis

1005 (A) Schematic illustration of the sequential sources of RNA pool over embryonic

1006 development. (B) Histogram shows the percent of differentially expressed genes in the Δm/wt

1007 versus f/wt embryos. Fold difference (in log2) is annotated as upregulated (with log2≥ 1;

1008 yellow), downregulated (as log2≤-1; green) and similar (as 1-1; grey). Number of

1009 genes is indicated on the right of the graph. Details concerning the RNA seq analysis are

1010 described in Materials and Methods section and Supplementary File 1 (C) Hierarchical

1011 clustering analysis for gene expression pattern of 16 libraries shows dramatic expression

1012 changes between f/wt (floxE1 to E8) and Δm/wt (ΔmE1 to E8) two-cell stage embryos. See

1013 also Figure 4-figure supplement 2 for analysis between two-cell stage and oocyte

1014 transcriptomes (D) RNA-seq data comparison with the different categories of the gene

1015 catalogue available at the Database of Transcriptome in Mouse Early Embryos (DBTMEE)

1016 generated an the ultralarge-scale transcriptome analysis (Park et al., 2013). The total

1017 number of genes belongings to each class and found in our RNA seq is indicated on top of

1018 the graph (see also Table 1). (E) Graphical representation of the normalized mean

1019 expression levels ± sem for chromatin-encoding genes in f/wt (in white) or Δm/wt (in black)

1020 MII oocytes (Oo, n=7) and two-cell stage embryos (2C, n=10). *corresponds to p<0.05 and **

1021 to p<0.001. (F) Top 6 representative GO terms (biological functions) enriched in Δm/wt

1022 mutant embryos. Fold overrepresentation indicates the percentage of misregulated genes in

41

1023 a particular category over the percentage expected on the basis of all GO-annotated genes

1024 present within the sequencing. p-value indicates the significance of the enrichment.

1025

1026 FIGURE 5. Increased LINE-1 protein levels and γH2AX foci in two-cell embryos

1027 depleted for KDM1A

1028

1029 (A) Pie chart representing the percent of each category of repeats analyzed in our 16 RNA-

1030 seq data of individual embryos. (B) Box-plot for percent of LINEs, SINEs and LTR element

1031 expression for f/wt (white) or Δm/wt (black) embryos over the total of reads mapping repeats

1032 for each of our 16 samples of RNA-seq. Details of the analysis are in experimental analysis.

1033 (C) qPCR analysis for LINE-1, SinesB1 and MuERV-L expression levels from individual two-

1034 cell stage cDNAs of f/wt (white) or Δm/wt (black). Each embryo is represented as a single

1035 bar. Data are expressed as normalized expression to three house-keeping genes. On the

1036 right of each graph is represented the mean± sem Two asterisks indicate p<0.01 as

1037 calculated using a Student’s t-test. (D) Nascent LINE-1 transcripts are detected by RNA

1038 FISH (signal in red) using a TCN7 probe on f/wt or Δm/wt two-cell stage embryos. RNAse A

1039 treated control embryos processed in parallel display no signal for RNA transcription. (see

1040 Figure 5-figure supplement 1 for Atrx expression control). (E) Quantification of LINE-1 RNA

1041 FISH. On the left, the graph represents the mean intensity of fluorescence(x axis) plotted

1042 against the respective maximum intensity (y axis) within each nucleus, and displays the

1043 distribution of RNA FISH signals in each nucleus of the two–cell embryos for the two

1044 populations (white squares= f/wt controls and black square= Δm/wt mutants. The two

1045 populations are significantly different by a t-test. On the right, box-plot representation of the

1046 entropy levels analysis of the RNA FISH images for the control versus mutant embryos as

1047 defined by Haralick parameters measuring the pattern of the image, here the changes of

1048 signal dots distribution. On the right, fluorescence quantification (one dot equals a f/wt (white)

1049 or Δm/wt (black) nucleus) where the mean fluorescence intensity is plotted against its

42

1050 respective maximum intensity and box-plot representation of the entropy levels analysis of

1051 the RNA FISH images for the control versus mutant embryo. P value is calculated with a

1052 student T test. (F) IF of two-cell stage embryos using anti-ORF1 antibodies (in red). A dotted

1053 line indicates the nucleus. Below is the graphical representation of the percentage of

1054 embryos displaying enriched fluorescent signal in either the cytoplasm (cy) or the nucleus

1055 (nu) for f/wt or Δm/wt embryos. (G) IF of two-cell stage embryos using antibodies directed

1056 against phosphorylated histone H2A variant X (γH2AX, in green) for f/wt and Δm/wt. Below is

1057 the corresponding quantification of embryo percentage according to the strength of γH2AX

1058 staining. DNA is counterstained by DAPI (blue). Number of processed embryos is indicated.

1059 Scale bar, 2μm (D, G)) 10μm (F).

1060

1061

43

1062 Table

1063

up down similar not in our data total in DBTMEE maternal 360 63 521 31 975 minor ZGA 540 47 750 40 1377 1C transient 34 4 51 1 90 major ZGA 10 297 297 10 584 2C transient 8 156 156 12 329 MGA 13 286 286 13 563 1064

1065 Table 1: Comparing the two-cell stage transcriptome of the Kdm1a mutant embryos to

1066 DBTMEE

1067 Numbers of genes found for the comparison of our two-cell stage RNA-seq data with the

1068 different categories for the gene catalogue found in DBTMEE. Total genes considered= 3811

1069 and total genes changed = 1818 (48%). Our dataset cover the genes categorized on the

1070 public resource with a minimum of 96% of genes.

1071

1072

44

1073 SUPPLEMENTS 1074 Figure 1-figure supplement 1: Kdm1a loss of function in female germline. 1075 (A) qPCR analysis of Kdm1a transcripts level from single ovulated oocytes cDNAs. Data are 1076 expressed as normalized mean expression levels ± sem for control (in white, n=7) versus 1077 mutant (in black, n=7) oocytes. Two asterisks indicate p<0.01 as calculated using a Student’s 1078 t-test. (B) IF analysis of control MII oocytes (from Kdm1a f/f females; left) or KDM1A depleted 1079 oocytes (from Kdm1af/f::Zp3cre females; right) with TUBULIN (green), and metaphase 1080 chromosomes are counterstained with DAPI (white). (C) Chromosome instability studied as a 1081 proportion of matured MII stages oocytes that exhibited normal alignment of chromosomes 1082 on the spindle versus matured MII stages oocytes with lagging chromosomes are shown in 1083 the graph. Oocyte genotypes are Kdm1af/f (in white; n=75) and Kdm1af/f::Zp3cre (in black; 1084 n=55). (D) Proportions of two-cell stage embryos with micronuclei: f/wt control embryos are 1085 shown in white n=60, compared to m/wt mutant in black n=40 (C and D). Data are 1086 presented as the mean ± s.e.d of two or three independent experiments. Statistical difference 1087 was calculated using a chi-square test. 1088 1089 Figure 1-figure supplement 2: Embryo recovery at day 2 post fertilization using natural 1090 matings. 1091 Distribution of developmental stages found in f/wt and Δm/wt embryos collected at embryonic 1092 day 2 (E2; expected two-cell stage) using natural breeding for Kdm1a f/f or Kdm1a f/f::Zp3cre 1093 females. Oocytes/embryos numbers, as well as the number of female per genotype are 1094 shown under the graph. 1095 1096 Figure 3-figure supplement 1: Immunofluorescence analysis of histone tail 1097 modifications upon maternal depletion or upon chemical inhibition of KDM1A for two- 1098 cell stage embryos. 1099 (A and B) Analysis of heterochromatin marks by IF with antibodies against me3 of H3K27 (A) 1100 and H4K20 (B) were performed on f/wt control embryos (left panel) in parallel to m/wt 1101 mutant embryos (right panel). (C and D) Analysis by IF with antibodies against H3K4me3 (C) 1102 and H3K9me3 (D) of two-cell stage wild-type embryos cultured in mock conditions (left 1103 panel) or with pargyline (right panel). Antibody staining is shown in green or red and DNA is 1104 counterstained with DAPI (blue). The number of embryos processed is indicated under each 1105 picture. Shown are full projections of stack sections taken every 0.5m. Scale bar, 10m. 1106 Under each image is the graphical representation of the fluorescent mean intensity ± sem 1107 (arbitrary unit, AU) of m/wt (in black) relative to f/wt (in white) two-cell stage immunostained 1108 embryos. 1109

45

1110 Figure 4-figure supplement 1: Immunostainings and RTqPCR analysis for assessing 1111 transcription of Kdm1a mutant two-cell stage embryos 1112 (A, B). Example of IF using anti-PolIICTD antibodies (A) or anti-PolII Ser2P (B) for f/wt 1113 control and m/wt mutant two-cell stage embryos with below the mean ± s.e.m fluorescence 1114 graphical representation (AU) of m/wt mutant embryos (in black) relative to f/wt control 1115 embryos (in white). Number of processed embryos is indicated under each image. Antibody 1116 signal in red, DAPI is blue. Scale bar, 10m. 1117 (C). Graphical representation of qPCR analysis for individual oocyte or two-cell stage 1118 embryos (f/wt control in white, n=10 or m/wt mutant in black, n=11) plotting the mean 1119 expression levels ± sem of three housekeeping genes Gapdh, Hprt and Ppia (according to 1120 GeNorm application). 1121 1122 Figure 4-figure supplement 2: Transcriptome analysis of Kdm1a mutant versus control 1123 in oocytes or two-cell stage embryos. 1124 (A) Venn diagrams show the numbers of differentially expressed genes in absence of 1125 KDM1A, when considering a fold difference (in log2) annotated as upregulated (with log2≥ 1) 1126 or downregulated (with log2≤-1) for two-cell stage embryos or ovulated oocytes (B) Principal 1127 component analysis for gene expression pattern of all libraries shows dramatic expression 1128 changes between f/wt and Δm/wt two- cell stage embryos, as well as with oocytes from both 1129 genotypes. (C) Venn diagram between oocyte stage or two-cell embryos for upregulated 1130 genes or down regulated genes (when comparing mutants to controls). (D) Top 6 1131 representative GO terms (biological functions) enriched in oocytes maternally deficient for 1132 KDM1A. Fold overrepresentation indicates the percentage of misregulated genes in a 1133 particular category over the percentage expected on the basis of all GO-annotated genes 1134 present within the sequencing. p-value indicates the significance of the enrichment. 1135 1136 1137 Figure 5-figure supplement 1: RNA FISH controls for LINE-1 ongoing transcription. 1138 RNA FISH using a probe recognizing the Atrx genomic locus (signal in red; DAPI is blue) on 1139 two-cell stage f/wt control embryo (left) and m/wt mutant embryo (right). Numbers of 1140 embryo processed and percentage of nuclei showing pinpoints of nascent transcripts by RNA 1141 FISH assays are indicated under each genotype. 1142 1143 Figure 5-figure supplement 2: EdU labeling and H2AX immunofluorescence of 1144 Kdm1a mutant versus control two-cell stage embryos. 1145 (A) Two-cell stage embryos (f/wt and m/wt) were collected at 40-41hr post hCG and culture 1146 for EdU pulse for 45 minutes. After fixation, they were analyzed by Click -It reaction to reveal

46

1147 the incorporated EdU (shown in green), then subjected to immunostaining for H2AX (shown 1148 in red). DNA is counterstained with DAPI (in blue). Shown are maximal projection of confocal 1149 z series of representative embryos. f/wt control embryo n= 17 and m/wt mutant embryo 1150 n=11. Scale bar, 10μm. 1151 (B) quantification of embryo percentage according to the presence of EdU incorporation or 1152 γH2AX immunofluorescence. 1153 1154 Rich media files 1155 1156 Supplementary File 1: Summary of RNA seq data for control and maternally depleted 1157 oocytes or two-cell stage embryos. 1158 Statistical analysis of all the single oocyte or individual two-cell stage embryo RNA-seq used 1159 in this study. Datasets are available from GEO under access number GSE75054 and 1160 GSE68139. 1161 1162 1163 Supplementary File 2: Differential Gene Expression at the two-cell stage and oocyte 1164 stage upon loss of maternal KDM1A. 1165 The DESeq R package was used to normalize and identify the differentially expressed genes 1166 between control and mutant embryos. Genes with 0 counts in all samples were filtered out 1167 and only the 60% of the top expressed genes were used for the analysis Differentially 1168 expressed genes were identified using a minimum Log2>1 (upregulation) or <-1 1169 (downregulation) fold change (FC) and with an adjusted p-value lower than α=0.05. 1170

47

FIGURE 1

A zygote two-cell stage B f/wt Δm/wt f/wt Δm/wt αKDM1A wb ponceau pb

m pb m 12 1 2

DAPI

DAPI pb + + 250 150 p p 100

75

KDM1A KDM1A

50

37

KDM1A KDM1A

C D oocytes f/wt Δm/wt IF stainings +24h +24h f/f cDNA qPCR f/f cre E2 E2 Kdm1a Kdm1a ::Zp3 100 culture culture =Δm 19 sperm 80

s

wild type o

y 19

r

b 60 75

zygote stage m

e 94 96 fragmented

f

IF stainings o 40 46 MII oocyte

% zygote 20 two-cell 15 three-cell four to/or Δm/wt 10 6 16 eight cell Kdm1a f/wt Kdm1a 0 4 two-cell stage n= 9 9 12 10 females n= 226170 206 53 oocytes/embryos IF stainings & culture RNA FISH E cDNA -RNA seq E2 +24h culture +48h culture -qPCR

control mutants embryos f/wt embryos (absence of the =100% maternal pool) of progeny =100% of progeny m/wt Δ

F Pargyline inhibition effect on in vitro zygotic development

Group No of zygotes No of embryos per stage (%) after 48hr in vitro used fragmented 2-cell 3/4-cell 8-cell Mock 74 0 3 (4%) 1 (1%) 70 (94%) a Pargyline 59 4 (7%) 45 (76%) 10 (17%) 0

a: none develop further when kept longer in culture A B

percent of embryos 100 20 40 60 80 FIGURE 2 0 aenlpoulu maternalpronucleus paternal pronucleus 99 f/wt me1

31 311 13 11 13 H3K4me3 DAPI H3K4me2 DAPI H3K4me1 DAPI pb p p p p p me2 p 088 8 10

me3 zygoticH3K9melevels zygotic H3K4melevels f/wt f/wt yoi 34elvl zygotic H3K9melevels zygotic H3K4melevels f/wt 77

' me1 m/wt m me2 m m m pb m me3 m IFstainingstrength: p p p p ' ' me1 ' m/wt m/wt f/wt p p m/wt me2 10 m me3 m m m m m ' me1 m/wt me2 me3 osann oeaestrong moderate no staining percent of embryos 100 20 40 60 80 0 D C 111 11 aenlpoulu maternalpronucleus paternal pronucleus

f/wt me1 181 8 11 8 11 me2 H3K9me3 DAPI H3K9me2 DAPI H3K9me1 DAPI 31 15 15 23 p p p p me3 p p f/wt f/wt f/wt ' 77 me1 m/wt m

me2 m mp m pb m m pb me3 pb p p p p me1 p f/wt ' ' ' m/wt m/wt me2 m/wt 23 me3 m m

' me1 m m m m m/wt me2 me3 FIGURE 3

A f/wt Δm/wt f/wt Δm/wt f/wt Δm/wt DAPI DAPI DAPI H3K4me3 H3K4me2 H3K4me1 H3K4me3 H3K4me2 H3K4me1 n=6 n=5 n=12 n=8 n=17 n=9 3 ** 3 9 *** 2 f/wt 2 6 Δm/wt 2.2x 1.9x 6.3x 1 1 3 relative to control relative to control relative to control fluorescence (AU) fluorescence (AU) fluorescence (AU) 1

B f/wt Δm/wt f/wt Δm/wt f/wt Δm/wt DAPI DAPI DAPI H3K9me2 H3K9me3 H3K9me1 H3K9me2 H3K9me3 H3K9me1 n=11 n=6 n=11 n=7 n=12 n=8 2 2 3 * * ** f/wt 2 Δm/wt 1 1 2.2x 1.2x 1.5x 1 relative to control fluorescence (AU) relative to control relative to control fluorescence (AU) fluorescence (AU) FIGURE 4

A fertilized oocyte zygote stage two-cell stage four-cell stage

mid preimplantation maternal pre-major major wave 'mE7 RNA transcription wave= =MGA stock wave= major ZGA 'mE6 minor Zigotic Gene Activation 'mE5 (ZGA) 'mE4 B C 'mE3 gene misregulation histogram 'mE2 'mE1 100 'mE8 = 2449 (up) 80 floxE7 = 2749 (down) floxE6 60 floxE4 floxE1

% of genes 40 floxE8 = 6167 (similar) 20 floxE5 floxE2 0 floxE3

D classified according to DBTMEE E

4 f/wt 'm/wt n=975 n=1377 n=90 n=584 n=329 n=563 * 100 ** ** * 80 3

60 ** ** ** **

40 2 gene expression % of genes 20

0 1

MGA similar normalized maternal transient down 0 Oo 2C Oo 2C Oo 2C Oo 2C minor ZGA major ZGA2C up not in our data H2az Suv39h1 Klf4 Atrx zygote transient

F gene ontology pathway analysis

Upregulated (maternal, minor ZGA and Zyg transient) downregulated (major ZGA, 2C transient; MGA) pathway termp value enrichment pathway termp value enrichment -6 -8 intracellular protein transport 2,7 10 2,9 ribosomal small 7 10 16 -11 subunit biogenesis protein transport 4,6 10 2,7 -5 -11 ribosome assembly 3 10 15 establisment of protein localization 6,5 10 2,6 -35 -7 ribosome biogenesis 4 10 12 cellular protein localization 1,5 10 2,5 -20 -7 rRNA processing 2 10 11 cellular macromolecule localization 1,9 10 2,5 -19 -8 rRNA metabolic process 1 10 11 cell cycle 3 10 2,4 -37 translation 5 10 9 FIGURE 5 AB f/wt p=0.018 ns ns 18,7% 25 80 8,8% 67,3% 15

60

Δ m/wt 15 16,4 10 12,9% 62,6% 40

% of normailzed repeats LINEs SINEs LTR LTR DNA others 5 5 20 LINEs simple repeat f/wt Δm/wt f/wt Δm/wt f/wt Δm/wt SINEs satellite

C LINE-1 SINE-B1 MuERV-L 8 f/wt Δm/wt

6 6 3

4 2 3 **

2 1 relative repeat expression

0 0 0 n=17 n=18 mean n=10 n=10 mean n=10 n=10 mean

D E p<0.001

4500 s

f/wt Δm/wt f/wt +RNase e 8

u

l

a

v 1

-

y

p E 3500 o r 7

t N Intensity

n

I DAPI

e

L

f

o

n

o

2500 i

t 6

u maximum

b

i

r

t

s

i

d LINE-1 p<0.001 1500 5 n=4 500 700 900 1100 f/wt Δm/wt n=23 n=21 mean Intensity

F f/wt Δm/wt G f/wt Δm/wt

X I

A P

2

A DAPI

H

D

γ ORF1 H2AX γ

ORF1 n=34 n=23

100 n=43 n=23 22% s 80

o

y IF staining r 17% 80 ~ 4 fold incidence b 60 strength:

m

e 82%

f absent

o

40 53% low 40 % 20 moderate 52 7 30 15% strong %of embryos 14 0 0 cynu cy nu f/wt Δm/wt f/wt Δm/wt